Machine Learning in Citizen Science: Promises and Implications

Julián Vicens,Laure Kloetzer,Marisa Ponti,Martina Franzen,Jakub Trojan

doi:10.1007/978-3-030-58278-4_10

Abstract

AbstractThe chapter gives an account of both opportunities and challenges of human–machine collaboration in citizen science. In the age of big data, scientists are facing the overwhelming task of analysing massive amounts of data, and machine learning techniques are becoming a possible solution. Human and artificial intelligence can be recombined in citizen science in numerous ways. For example, citizen scientists can be involved in training machine learning algorithms in such a way that they perform certain tasks such as image recognition. To illustrate the possible applications in different areas, we discuss example projects of human–machine cooperation with regard to their underlying concepts of learning. The use of machine learning techniques creates lots of opportunities, such as reducing the time of classification and scaling expert decision-making to large data sets. However, algorithms often remain black boxes and data biases are not visible at first glance. Addressing the lack of transparency both in terms of machine action and in handling user-generated data, the chapter discusses how machine learning is actually compatible with the idea of active citizenship and what conditions need to be met in order to move forward – both in citizen science and beyond.

Highlights

The combination of human and machine learning, wherever they complement one another, has a lot of potential applications in citizen science
We address two main questions here: (1) what tasks are citizens being invited to perform in citizen science projects through the use of machine learning (ML)? and (2) what are the main risks and opportunities of using ML in citizen science? The majority of citizen science projects are centred around data provided, for example, by satellites, cameras, or, more generally, sensors (Neal 2013)
The processing power and sophistication of algorithms have improved at previously unimaginable levels, and some ML techniques have already outperformed or at least parallelled human capabilities

Summary

Introduction

The combination of human and machine learning, wherever they complement one another, has a lot of potential applications in citizen science. Several projects have already integrated both forms of learning to perform data-centred tasks (Willi et al 2019; Sullivan et al 2018). ML algorithms are currently the most widely used and applied, for example, in image and speech recognition, fraud detection, and reproducing human abilities in playing Go or driving cars. In scientific research, they find many applications in different fields such as biology, astronomy, and social sciences, just to mention a few (Jordan and Mitchell 2015). The Galaxy Zoo project and the classification and identification of galaxy morphological shapes is a good case in point

10 Machine Learning in Citizen Science

Future Trends, Recommendations, and Conclusions